Estimates of the severity of coronavirus disease 2019 (COVID-19) using model-based analysis showed a mean time from symptom onset to death of roughly 18 days, a mean length of hospital stay of approximately 25 days, and a case fatality ratio of 1.38%. A strong age gradient in mortality risk was also seen, with patients aged >80 years having a fatality ratio of 13.4% and infection fatality ratios significantly increase at age >50 years, according to data published in Lancet Infectious Diseases.
Thus far, fatality ratio estimates for COVID-19 have substantially differed in magnitude. Estimates using data on hospitalized patients reported the highest fatality ratios (8%-28%), whereas population level estimates across all age groups ranged from 0.2% to 1.6%. The study was therefore designed to synthesize data across a variety of surveillance settings to obtain estimates that account for differing denominator populations in the datasets.
Investigators took individual case data for 24 patients with COVID-19 in China, for whom all necessary data elements were available to estimate time from symptom onset to death. To estimate symptom onset to recovery time, investigators used case data from 169 recovered patients outside of mainland China. To obtain case fatality ratio estimates, researchers examined case data on 1334 international patients.
With adjustments, the posterior estimate of mean time from symptom onset to death from COVID-19 using the aggregated data was 17.8 days (range, 16.9-19.2). Based on data from the 169 individuals outside of mainland China who recovered, the estimated mean time from symptom onset to recovery was 24.7 days (95% credible interval [CrI], 22.9-28.1) with a coefficient of variation of 0.35 (0.31-0.39); these data are consistent with a previously reported study from China. Accounting for under-ascertainment, the highest fatality ratio was estimated in the age group of >80 years at 13.4% (95% CrI, 11.2-15.9), and the lowest was in the age group of 0 to 9 years at 0.00260% (95% CrI, 0.000312-0.0382).
For the 1334 international cases, investigators used parametric modeling to estimate an overall modal case fatality ratio of 2.7% (95% CrI, 1.4-4.7). For individuals who reported travel to mainland China, the estimated overall model case fatality ratio was 1.1% (95% CrI, 0.4-4.1), and in those who did not travel to China the ratio was 3.6% (95% CrI, 1.9-7.2). When stratified by age, the case fatality ratio was lower for individuals aged <60 years (1.4%; 95% CrI, 0.4-3.5) compared with those aged >60 years (4.5%; 95%CrI, 1.8-11.1).
Using data from 689 repatriated international Wuhan residents, infection prevalence was estimated to be 0.87% (95% CrI, 0.32-1.9). Adjusting for demographics and under-ascertainment, the infection fatality ratio was 0.66% (95% CI 0.39-1.33). The case fatality ratio estimates rose steeply from age >50 years. The percentage of individuals with the virus requiring hospitalization ranged from 1.1% in individuals aged 20 to 29 year to 18.4% in those aged >80 years.
Using these findings, investigators also estimated the population infection fatality ratio of a quarantined cruise ship, the Diamond Princess, to be 2.9%. They anticipated a 0.66% infection fatality ratio that did prove to be consistent with that observed up to March 5, 2020, but was slightly over the 95% CI age-adjusted limit observed by March 25 (of 712 passengers with COVID-19, 10 have died, 11 remain in critical condition, and 601 have been discharged). The differences in estimated and actual numbers may be attributable in part to the passengers receiving better than average quality health care.
Study investigators concluded, “Our estimates of the underlying infection fatality ratio of this virus will inform assessments of health effects likely to be experienced in different countries, and thus decisions around appropriate mitigation policies to be adopted.”
Disclosure: Several study authors declared affiliations with the pharmaceutical industry. Please see the original reference for a full list of authors’ disclosures.
Verity R, Okell LC, Dorigatti I, et al. Estimates of the severity of coronavirus disease 2019: a model-based analysis [published online March 30, 2020]. Lancet Infect Dis. doi:10.1016/S1473-3099(20)30243-7